AN ANALYSIS OF COI'IPOSITIONAL AND MICROSTRUCTURAL EFFECTS
ON THE RESISTANCE OF A PROTOTYPE SPARK PLUG
RESISTOR MATERIAL
A THESIS
Presented to
The Faculty of the Division of Graduate Studies
byJack Howard Logan, Jr.
In Partial Fulfillment
of the Requirements for the Degree
Master of Science
in the School of Ceramic Engineering
Georgia Institute of Technology
June, 197 8
AN ANALYSIS OF COMPOSITIONAL Al^D MICRO STRUCT URAL EFFECTS
ON THE RESISTANCE OF A PROTOTYPE SPARK PLUG
RESISTOR MATERIAL
Approv.ed •-
A i a n T. (jnapman, y n a i r m a i r
J o s e p h ' L.
Pentecost
Jofe K. C o c h r a n ,
Jfi^
D a t e a p p r o v e d by C h a i r m a n : JjjrJ/ltt,
^/^jfci
ii
ACKNOWLEDGEiMENTS
Dr. A. T. Chapman v/as my thesis advisor and chairman
of the reading committee, and I would like to thank him for
his advice, support, and assistance throughout the scope of
this thesis.
Thanks go to Dr. J. K. Cochran for
his advice
in obtaining the photomicrographs and to Dr. J. L. Pentecost
for his comments on the structural model.
These two men also
served on the reading committee.
Mr. Tom Machrovitch deserves a special word of appreciation for his useful technical knowledge and ideas for solving
problems, and for his constant good humor.
I would like to thank the Prestolite Division of the
Eltra Corporation for providing the financial assistance for
this project.
Mr. B. D. Fitts was very helpful with his
background and production information on spark plugs, and I
particularly appreciate his willingness to offer assistance
throughout the research.
I also thank the following people for the important
contributions they provided in their respective ways:
Mary
Ann Breazeale, Peggy Castro, Jim Price, and Debbie Welsh.
I l l
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS
LIST OF TABLES
LIST OF ILLUSTRATIONS
SUMMARY
ii
V
vi
viii
Chapter
I.
II.
INTRODUCTION
1
SURVEY OF LITERATURE
3
Purpose and Design of Resistor Spark Plugs
Review of Patent Literature
Properties of Resistor Mix Components
III.
MATERIALS, PROCEDURES AND EQUIPMENT
27
Components of the Resistor Mix
Mixing the Components and Loading the
Insulators
Sealing Apparatus and Procedure
Resistance Values
Photomicrographs
IV.
RESULTS AND DISCUSSION
40
Introduction
N, e, and Q
Resistor Mix ID
P, R, and S
Resistor Mix ID
Resistor Mix ID
T, U, and V
Microstructural Examination of the Resistor
Component and the Copper-Glass
Model for Current Flow Through the Resistor
Mix
Discussion of the Model
V.
CONCLUSIONS
64
IV
Page
APPENDIX
A.
B.
C.
D.
HEAT TREATMENT INFORMATION DOW CORNING
RELEASE COATING Ql-2531
67
A DESCRIPTION OF THE SPARK PLUG INSULATOR,
ELECTRODES AND COPPER GLASS
71
ATTEMPTS TO REDUCE THE SCATTER IN THE SPARK
PLUG VALUES
73
A CALCULATION FOR THE VOLUME OF TiO
THE RESISTOR MIX
80
BIBLIOGRAPHY
IN
81
V
LIST OF TABLES
Table
1.
2.
3.
4.
5.
6.
7.
8.
9.
•
Page
The Electrical Resistivity at Room Temperature
and the Temperature Coefficient of Resistivity
of TiO^ Samples Fired in Nitrogen-^'
18
The Electrical Resistivity (J2-cm) of Rutile as
a Function of Sintering Atmosphere and Temperaturel8
18
The Resistivity as a Function of Small Values
of X in TiO^
for Rutile Samples Heated to
19
1000°C at Various Oxygen Pressures
22
The Resistivity of Flux Grown Titania Oxides as
a Function of Large Values of X in TiO^
at
27°c20
^7^. . . .
22
The Compositions and Resulting Spark Plug
Resistances for Two Component Resistor Mixes,
N, e, and Q
41
The Compositions and Resulting Spark Plug
Resistance for Three Component Resistor Mixes:
P, R, and S
44
The Compositions and Resulting Spark Plug
Resistance for Four Component Resistor Mixes:
T, U, and V
48
The Lack of Correlation Between Resistance
Values and the Gap Distance
74
The Composition of the Resistor Mixes and the
Resistance Values of the Spark Plugs Incorporating These Mixes
76
'MHlMMBaMliiWiWUMHWIpWIllMtll
VI
LIST OF ILLUSTRATIONS
Figure
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
Page
A Comparison of the Voltage Oscillations of
Resistor and Non-Resistor Spark Plugs^
5
A Schematic Drawing Displaying the Internal
Parts of a Non-Resistor Spark Plug
6
A Schematic Drawing Showing the Resistor
Component and Copper-Glass Seals of a Resistor
Spark Plug
"7
Electrical Resistivities of Sam.ples of TiO^
(Rutile) Heat Treated and Reduced in
Hydrogen^^^
Electrical Resistivities of Unsintered Samples
of TiO-5 (Rutile) at Oxygen Pressures of (A)
-4
-2
1.3 X 10
atm., (B) 6.6 x 10
atm., and
(c) 1.05 atm.-^^
21
The Electrical Resistivity of Boron Carbide
.
34
The Volume Resistivity of Corning 7070 Glass .
A Schematic Drawing of Sealing a Loaded
Resistor Spark Plug
19
21
24
29
31
The Apparatus Built to Seal Spark Plugs in a
Manner Typical of Production Techniques: (A)
General View, (B) Close-Up of the Stool and
Torches
33
The Circuitry Used to Modify the Millivolt
Recorder to Measure Resistance Values of the
Spark Plug Throughout the Sealing Processes . .
35
A Millivolt Recorder Plot Showing Spark Plug
Resistance Values Throughout the Sealing
Process
37
A. Picture Showing the Resistor Component,
Copper-Glass and Electrodes Sealed in an
Insulator
39
Vll
Figure
13.
14.
15.
16.
Page
The MiniitiLim Resistance Values Measured During
the Heating of Two Component Resistor Mixes:
N, 0, and Q
42
The Final Resistance Values as a Function of
Percent B,C at Two Ti02 Contents in a Resistor
Mix Composed of Glass, TiO^ and B.C
46
Resistance Values Vary with the Content of
Ti02/ B-C and Binder
50
Photomicrograph of a Sealed Resistor Component
Revealing that Laminar Flow of the Material
Occurred During Hot Pressing: (A) TiO^ Layers,
(B) Glass Layers, (C) B.C Particles (542x). . .
17.
Photomicrograph of a Sealed Spark Plug Showing
the Copper-Glass and the Resistor Component:
(A) Copper-Glass, (B) Ti02 Layers, (C) B^C
Particle, (D) Pores in the Glass-Ti02 Matrix,
(E) Copper-Glass Belov; the Surface of the
Sample, (F) Ti02 in the Pores (196x)
18.
19.
20.
21.
54
55
Photomicrograph Showing the Interface Between
the Sealed Resistor Component and the Center
Wire Head: (A) Center Wire Head, (B) B^C
Particles, (C) Pores in the Glass-Ti02 Matrix,
(D) Ti02 in the Pores (196x)
56
A Schematic Drawing of the Resistor Component
Shov/ing Possible Current Paths Through (A) The
Area Between the Electrodes and (B) The Area
Along the Edges of the Insulator Bore
60
The Weight Loss of Binder Due to Heat Treatment
at 982°C
69
The Spark Plug Electrodes and Insulator Showing
the Shoulder on Which the Center Wire Head Rests
in the Low^er Part of the Bore
72
Vlll
SUMT'IARY
This investigation considered the compositional
parameters of a spark plug resistor mix where the principle
components were a borosilicate glass, titania powder, boron
carbide particles, and a silicone resin binder.
Further, a
model was devised to explain the conduction of the current
through the resistor mix.
This was accomplished using com-
positional information in conjunction with a detailed microstructural analysis of the resistor mix.
The microstructural analysis revealed that the sealed
resistor mix had a laminar structure that was perpendicular
to the electrodes between the electrodes and parallel to the
electrodes along the bore wall of the insulator.. The TiO^
particles spread out in the layers betv/een the glass particles
and was the primary conductor in the resistor mix.
The B,C
particles served as conduction paths connecting some layers
of TiOp.
The binder content affected the resistance values
because it reduced the Ti02.
The anisotropy of the resistor
macrostructure effectively served to minimize variable resistance values caused by non-uniform dimensions of the various
components of the spark plugs.
CHAPTER I
INTRODUCTION
Ceramic resistors have been incorporated with spark
plugs since the advent of autoraobile radios.
The discharge
of high voltages across the spark plug gap generates radio
frequency interference.
One method of reducing the inter-
ference is to place a ceramic resistor in the bore of the
spark plug insulator in series between the electrodes.
There
is little, if any, information in the non-patent literature
on a high temperature ceramic resistor for spark plug application.
There are many characteristics which a resistor spark
plug should possess.
The two primary characteristics are that
the resistance value of the spark plug be sufficient to suppress the interference, and that the resistance value be maintained under engine operating conditions and during repeated
usage over long periods of time.
These conditions necessitate
the use of ceramic materials because conventional electric
resistor materials fail at high temperatures and high voltages
A resistor composition is typically glass and semiconductors
mixed with a binder.
The resistor mix is placed in the bore
of the spark plug insulator between the electrodes and the
assembly is heat treated to seal the resistor mix and the
electrodes in the insulator.
Many parameters affect the final resistance value
created by a resistor mix in a spark plug.
The purpose of
this investigation was to study the compositional parameters
of a resistor mix where the principle components were a
borosilicate glass, titania pov7der, boron carbide particles,
and a silicone resin binder.
The selection of these com-
ponents was based on a prototype composition and on evaluating
many compositions reported in the patent literature, and the
effects of each component on the spark plug resistance were
studied.
A second objective of the investigation was to
devise a model which explained the conduction of the current
through the resistor mix.
This was accomplished using the
compositional information in' conjunction v/ith a detailed
microstructural analysis of the sealed resistor mix.
It is
hoped that the results of this investigation will advance
the understanding of ceramic resistors for spark plug application and solve some of the problems associated with the
production of resistor spark plugs.
•MUMliitHllrMlllllllK
CHAPTER II
SURVEY OF LITERATURE
This literature survey is a review of the state of the
art of ceramic resistors as incorporated in spark plugs.
A
brief introduction is given on the purpose and design of
resistor spark plugs, and the composition and desirable
characteristics of the resistor mix.
Most of the information
on the state of the art is contained in patents, and therefore details are limited.
The pertinent characteristics are
discussed of each resistor mix component studied in this
research.
Purpose and Design of Resistor Spark Plugs
The purpose of the resistor spark plug is to damp out
radio frequency oscillations which occur when the spark discharges.
The oscillations cause automobile radio frequency
interference particularly in AM and shortwave reception, and
1 2
they cause rapid erosion of spark plug electrodes. '
They
also may interfere with other electronic equipment in close
3
4
proximity to the automobile.
Burgett, et. al. explains
that a spark plug acts as a capacitor which discharges when
the spark occurs, generating interference.
A resistor spark
plug suppresses the interference by dividing the capacitive
effect and reducing the radio frequency interference.
The
voltage discharge oscillations of resistor and non-resistor
spark plugs are compared in Figure 1.
In the bore of a spark plug, the lower electrode where
the spark occurs is called the center wire and the upper electrode is called the terminal screw, as shown in Figure 2.
Standard spark plugs have the two electrodes connected and
sealed in place by an electrically conducting metal-glass
seal.
Resistor spark plugs differ in that the glassy seal
incorporates a ceramic resistor mix.
Most ceramic resistor mixes do not bond well to the
electrodes, but they do bond well to metal-glass seals similar
to those used in standard spark plugs.
In addition, some
of the components that may be used in the resistor mix are
organic and subject to oxidation at spark plug operating
3
.
.
.
temperatures.
For these two reasons, the resistor mix is
often positioned between two layers of a metal-glass composition, as shown in Figure 3.
The metal-glass composition,
typically copper-glass, serves as a seal to protect the
resistor mix from oxidation and it bonds well to the resistor
mix, the insulator and both electrodes.
is called three-step loading.
This layering process
If adequate bonding can be
achieved between the resistor mix and the electrodes, one or
both copper-glass seals may be eliminated in which case the
resistor mix also serves as the sealing and bonding mechanism.
In assembling the resistor spark plug, the center wire
is placed in the insulator and is usually supported by a
•Non-Resistor
Spark Plug
•Resistor
Spark Plug
Time
Figure 1.
A Comparison of the Voltage Oscillations
I
of Resistor and Non-Resistor Spark Plugs
B*«i«ii>iwwiiiMi|M|«gM8B,|g|||MMMg|fc
terminal
screw
insulator
coaxial
grooves
copper-glass
seal
center wire
head
center wire
Figure 2. A Schematic Drawing Displaying the Internal
Parts of a Non-Resistor Spark Plug.
copper-glass
seals
Figure 3.
resistor
component
A Schematic Drawing Showing the Resistor
Component and Copper-Glass Seals of a
Resistor Spark Plug.
shoulder in the bore.
A lower metal-glass powder, the
resistor mix, and an upper metal-glass powder are placed in
the bore on top of the center wire.
These components are
pre-pressed, either individually or as a group, to increase
the density of the final, sealed resistor mix.
The terminal
screw is positioned on top, protruding above what is to be
its final position.
This assembly is fired to a temperature
sufficient to soften all the glass and the terminal screw is
hot-pressed into the bore.
The pressure causes the molten,
conductive metal-glass to fill all space around the head of
the center wire and the lower end of the terminal screw.
resistor mix remains between the metal-glass seals.
The
The
assembly is allov/ed to cool under pressure until all components are firmly bonded in place.
The result is a spark
plug as shown in Figure 3 with electrodes tightly sealed to
the insulator and to the resistor mix by the metal-glass
seals.
In the cases where one or both metal-glass seals are
not required, a similar procedure is used without the unneeded
seal (s).
The resistor mix is basically a glass containing
semiconductor materials.
There may also be inert fillers,
metal powders, carbonates, other carbonaceous materials,
reducing agents, and/or a binder.
These components, v;hen
mixed thoroughly and sealed by heating in the bore of the
spark plug, create a dense, low porosity mass.
The result
is a series of semiconductor particles separated by glass
and other non-conducting components.
The ignition current
jumps from particle to particle as it rfioves through the
resistor mix.
The spark plug resistance value is primarily
dependent on the characteristics and amounts of the components in the resistor mix, but other factors including
sealing temperature and furnace atmosphere also affect the
resistance values.
The manufactured resistor spark plug must have a given
resistance value, usually around 10 kQ, and low scatter of
the resistance values is desirable.
In operation, the
sealed resistor mix, which serves as all or part of the seal
between the electrodes, has to be strong to prevent the
electrodes from becoming loose, and impervious to engine
cylinder gases under pressure.
The resistance value should
have low temperature and voltage coefficients, and it should
exhibit little electrical aging (change in electrical properties
with time, particularly due to applied voltage).
All these
characteristics must be maintained up to 540°C. 7
Review of the Patent Literature
There are three broad categories of resistor mixes
for use in spark plugs.
One category uses a semiconductor,
such as titanium dioxide, and one or more reducing agents.
Another category includes the use of a carbonaceous material,
a reducing agent, and a metallic carbonate.
The final cate-
gory uses a ceramic filler and a tin oxide resistor component.
«««wi««n»w« 11 mmmm^timmiiifitKiiiit
10
In all cases a glass is used as a structure to hold the
other components.
A patent representative of the first category was
3
granted to Heischman in 1963. It discloses a resistor mix
consisting of a borosilicate glass, titanium dioxide and
boron carbide, all held together with a binder.
Corning
glass #7 07 0 is used because it has a low thermal expansion
~ fi
of 3.2 X 10
inches/inch/°C.
Titanium dioxide, pigment
grade, is easily reduceable to form suboxides which are semiconductors.
Boron carbide, 22 0 mesh, acts as a semiconduc-
tor and a reducing agent.
The binder is Methocell, 3000 cps.,
a water soluble cellulose gum, which also acts as a reducing
agent.
An important factor in Heischman's patent is that the
need for metal-glass seals on either side of the resistor
mix is eliminated.
There are two reasons for this.
The
components used are inorganic and are, therefore, relatively
stable at the operating temperatures and pressures of spark
plugs.
So, they do not need protection from oxidation.
Also, the electrodes are designed in a manner that
promotes bonding with the glassy resistor mix.
screw is threaded at its lower end.
The terminal
It also has a thermal
expansion greater than that of the resistor mix, due to the
low thermal expansion of the glass.
The terminal screw cools
after hot-pressing and contracts more than the resistor mix.
Tension is created between the terminal screw head on top and
•••iiniiMiiiirimiiiiiiiiiiiiiHI
the lower terminal screw grooves which are mechanically seated
in the resistor mix.
The center wire has a thermal expansion of 13 x 10
—6
inches/inch/°C and the head is shaped cylindrically, with
upper and lower flanges curved inwardly.
This design creates
circular concave lips which, upon cooling, firmly secure the
resistor mix to the center wire.
The center wire head also
has a raised middle area for more electrical contact with the
resistor mix, and there are vertical ridges on the outer part
of the center wire head to prevent turning.
The designs of
these electrodes result in good mechanical seals which
allow ample contact with the resistor mix, eliminating the
need for intermediate metal-glass seals.
An example of a composition from this patent (in weight
percent) is 88.0 w/o Corning #7070 glass, 9.5 w/o TiO^,
2.0 w/o B.C and 0.5 w/o Methocel.
value of 10 k^.
This gives a resistance
In all compositions, the amounts of Methocel
and B.C are held constant.
The resistance value is controlled
by adjusting the TiO^/glass ratio, which is inversely proportional to the resistance value.
For production, water is added to the mixed components,
the mix is dried, granulated, and loaded into the spark plug,
and the assembly is heated to 9 00*^C.
The terminal screv; is
pressed in and the assembly is allowed to cool to below red
heat, at which time the pressure is removed.
Heiscliman notes factors, other than the composition.
iiiliiiiUulJllllljMlilll^^
12
which affect the resistance value and the seal.
These include
the dimensions of the resistor mix after sealing, the furnace
atmosphere, whether reducing or oxidizing, and the sealing
temperature.
The patent granted to Webb, et. al., in 1971 is typical
of the second category of resistor mixes, and improves on
8 9
two earlier patents. '
This patent claims to satisfy the
desired requirements of resistor spark plugs with an extensive and complicated composition.
The glass is a barium
borate, the amount of which has little effect on the resistance values.
An inert filler such as kyanite, zirconia, or
mullite, etc., controls the fluidity of the mix during hot
pressing.
Bentonite is used as the binder to get the mix
into granular form for volumetric loading.
Organic binders
are not used because they cause the resistance values to be
unstable.
Carbon black is added as a reducing agent or a
conducting medium.
A water soluble, carbonaceous material
such as sucrose, Methocel, or corn flour, etc., is used to
combine with the carbon black and/or semiconductor material
to reduce the electrical aging of the resistor mix.
The Webb, et.al., patent reveals the use of certain
metallic carbonates and metal powders v/hich reduce the
porosity of the resistor mass and strengthen the bond
between the resistor mass and the center wire, thereby
elminating the need for a lower metal-glass seal.
The
carbonates are of lithium, zinc, sodium, or magnesium and
13
the metal powders are zinc, antimony or tellurium.
The
melting points of these range from 300 to 820°C; at a
sealing temperature of 870 to 930°C, these materials are
probably in a liquid form during hot pressing.
decomposes and gives off carbon dioxide.
The carbonate
The cation enters
the glass phase and modifies the viscosity of the glass in
order to inhibit the flow of the molten metal powder below
the center wire head.
This patent claims that the molten
metal coats the center wire head forming a strong bond
between the center wire head and the glassy phase of the
resistor mix.
This bond eliminates the need for a lower
metal-glass seal, but an upper seal is still needed.
The
modified glass is also less porous, thereby strengthening the
resistor mix.
Included in this patent is also the use of a semiconductor material for cases of extreme electrical requirements,
7 9
9
as described by Counts, et. al. '
In one case, the material
is a binary m.etal oxide system, particularly sintered titanium
zirconate.
Before incorporating the semiconductor material
in the resistor mix, the material is heated in the range of
1260°C to 1620°C for one to 24 hours, and cooled and ground to
7
-200 mesh. In the other case, the material is a stannous
titanate semiconductor and must be pre-reacted similar to the
above procedure.
In both cases, the resistance value is con-
trolled by the content of the semiconductor material and by
the carbon black, which is used to reduce the semiconductor
14
material.
An example of a resistor mix (in parts by weight, pbw)
consists of 0.5 pbw lithium carbonate, 5 pbw zinc, 30 pbw
barium borate glass, 44 pbw zirconia, 25 pbw Indian Kyanite,
1.8 pbw Thermax (carbon black), 0,34 pbw sucrose and 1.83
pbw bentonite.
The resistor mix size is -28, +100 mesh and
it is volumetrically loaded into the insulator.
The sealing
temperature is 915°C with an 18 minute heating period followed
by a 7 minute hold at 915°C.
A patent granted to Yoshida and Sakai,
from Japan,
in 1977 is similar to the Webb, et. al., patent, making use of
metals and carbonaceous materials for sealing, strength and
electrical stability.
An important difference is the Japanese
use no carbon black, stating that it causes unevenness in the
resistance values.
A carbonaceous material is the sole source
of carbon.
The third type of resistor mix is described in the
2
patent granted in 1977 to Oki, et.al.
A theory is proposed
where the radio frequency noise can be suppressed if the
electric current moving through the spark plug can be made
to "zigzag."
This is accomplished with a resistor mix con-
taining three basic materials, a resistor component, a
ceramic filler, and a soft glass.
The resistor component and ceramic filler work together
to create the zigzag path.
The resistor component is primarily
tin oxide, with small additions of carbon powder, antimony
15
oxide, tantalum oxide, aluminum phosphate and/or a binder.
The resistivity of tin oxide produces spark plug resistance
values from several k.Q to several tens of k^.
The ceramic
filler such as alumina^ zirconia, and quartz glass, etc. has
a resistivity much higher than the resistor component.
Since
the current does not flow through the filler but flows through
the resistor component, current "zigzags" through the resistor
mix.
The third component, a soft glass, may be of almost any
composition with a softening temperature between 300 and 600°C.
The softening temperature is above 300°C because engine operating temperatures are up to about 250°C.
This value is low
compared to other resistor mixes which are designed to operate
7
up to 54 0°C.
This patent proposes that spark plugs with
glasses that have softening temperatures above 6 00°C give high
noise levels when tested in automobiles.
This is because the
glasses are not fluid enough during the hot pressing to permit
the resistor mix to bond well to the bore of the insulator and
to the copper-glass seals.
In contrast to VJebb, et al.,
Oki, et al., says the amount of glass does affect the resistance values, but that the stability of the resistivity of
tin oxide considerably reduces this effect.
A satisfactory composition (in volume percent) contains
10 v/o glass, 70 v/o filler and 20 v/o resistor component.
There is no binder; the three components are just mixed in
a ball mill.
The composition of the glass (in weight percent)
16
is 11 w/o B O^/ 11 w/o AI2O-, 3 \v/o SiO^, and 7 5 w/o PbO.
It has a thermal expansion of 8.3 x 10
temperature of 44 0°C.
— f\
/°C and a softening
The filler is quartz glass which has
a softening temperature of 1650°C.
The resistor component
contains 87 w/o Sn02 / 10 w/o 'Jla.<:)^ and 3 w/o Sb^Oo? which are
calcined together at 1200°C before being mixed with the glass
and filler.
This composition gives a spark plug resistance
value of about 5 k^.
The resistance value is adjusted by
changing the amount of tin oxide in the resistor component.
The amounts of the other components, including the total
amount of the resistor component, remain constant.
Copper-glass seals are used on both sides of the resistor mix.
The copper-glass powders and the resistor mix are
loaded by weight and each is pressed into the bore with about
2
50 kg/cm as it is loaded. The assembly is heated in an
electric furnace at 350°C for 30 minutes.
It is removed from
2
the furnace and the terminal screw is pressed v;ith 50 kg/cm .
Properties of Resistor Mix Components
The selection of the components of the resistor mix
studied in this research was based on a prototype composition
and on the first category of resistor mix described above
The materials are titanium dioxide (anatase structure),
boron carbide, a borosilicate glass and a silicone resin
binder.
A discussion follows of the applicable characteristics
and properties of these materials.
The specifications of the
particular products that were used are given ].ater.
17
Electrical Characteristics of TiO^
Titanium dioxide is a semiconductor v/ith two structures,
anatase and rutile, which are stable at room temperature.
When
anatase is heated to between 3 00 and 1100°C, it changes to
rutile,
which structure remains when the sample is cooled
to room temperature.
The intrinsic resistivity of rutile 12
13
IS about 10
and 10
2
Q-cm at room temperature, 10
P.-cm at 1000°C.
f^-cm at 300°C,
Other values are reported
"" and
the discrepancies are probably due to specimen preparation
measurement techniques, or the oxidation condition.
The resistivity of rutile changes as a function of
stiochiometry because rutile is a reduction type semiconductor
which readily forms suboxides of TiO^__ .
The extent of the
oxygen loss strongly depends on the temperature and the
atmosphere in which the TiO^ is treated.
Table 1 presents
data for the effect of heat treatment on the electrical
resistivity of samiples that were pressed and then fired in
nitrogen.
Table 2 shows the effects of atmosphere and temper-
ature for sintered samples of TiO^ (rutile).
The effect of
hydrogen reduction on rutile ceramics is shown in Figure 4.
This figure is presented not only to show the effects of
tem.perature but also that the reduction time has little
effect, at least at 300°C.
Further, the resistivities of
the samples changes less than one order of magnitude v/hen the
temperature at which the resistivity is measured is varied
from -170°C to 727=C.
The slope- is positive between 20^C and
18
Table 1.
The Electrical Resistivity at Room Temperature and
the Temperature Coefficient^of Resistivity of TiO^
Samples Fired in Nitrogen,-^^
Resistivity
(J2--cm)
Temp. Coeff.
of Resistivity
(20-200°C)
950
4.34 X 1 0 ^
+1.2 X 10^
1000
1.42 X 10^
+3.1 X 10^
1200
1.9
+ 4.2 X lO"'-
Heat
Treatment (°C)
Table 2.
X 10*^
The Electrical Resistivity (fi-cm) of Rutile as a ,„
Function of Sintering Atmosphere and Temperature.
Sintering Temperature
Sintering
Atmosphere
N2
Air
0^
1300°C
6.4 X 10^
lO-^-^
11
10-^-*-
1400°C
6.0 X 10^
1.4 X 10^
12
10-^
19
700
•-100
100 0
175
-150
4-
-f10
e
u
I
B
O
— (C) (D)
>i
-P
•H
>
•H
-P
CD
•H
CD
Heat Treatment
Temp (QQ Time (m.in)
600
10
(A)
(B)
700
10
(C)
800
10
(D)
800
5
(E)
900
10
0.1
Q)
0.01
1 2
Fiqure 4
3
4
5
7 8
1/T°K X 10^
9
10 11 12 13
Electrical Resistivities of Samples of TiO^
(Rutile) Heat Treated and Reduced in Hydrogen-'-^^
20
200°C indicating a negative temperature coefficient of
resistivity in that range, as opposed to the positive
coefficients for non-reduced samples presented in Table 1.
Figure 5 shows the effect of oxygen pressure on resistivity.
The maximum value of x in TiO^
has been debated.
2-x
When reducing TiO^, but maintaining the rutile structure, it
may be as high as 0.01, but x can be much larger when
specific crystals are flux grown.
19 20
'
Tables 3 and 4 are
presented to give a quantitative idea of the change in
resistivity as a function of stoichiometry.
Table 3 was
developed from information on samples heated to high temperatures at various oxygen pressures to form non-stoichiometric
rutile.
Table 4 comes from information on flux grown single
crystals of titanium oxides tested at room temperature.
The quantitative effect of titania in a resistor mix
sealed in a spark plug is difficult to determine.
The infor-
mation in Tables 2, 3, and 4 and Figure 5 is on single
crystals or polycrystalline pellets whose density approaches
that of the theoretical value.
in a sealed resistor mix.
Neither condition is realized
The resistivity depends not only
on the inherent properties on the suboxide created by heating,
but also on the shape, orientation and packing of the poly14
crystalline conglomerates.
Qualitatively, under spark plug sealing conditions,
the extent of the reduction of the titania probably falls in
the area of the information in Table 3.
Even a slight reduc-
21
500°C
e
o
>i
-p
•H
>
•H
-P
U
P
13
G
O
u
en
O
i^
10
11
1/T^K
Figure 5.
12
13
X lO"^
Electrical Resistivities of Unsintered
Samples of Ti02 (Rutile) at Oxygen Pressures
of (A) 1.3 X 10"^ atiri., (B) 6.6 x 10"^ atm. ,
and (C) 1.05 atm."^^.
22
Table 3.
The Resistivity as a Function of Small Values of
X in TiO^
for Rutile Samples Heated to 1000°C
^
j^^
X
at Various Oxygen Pressures.
X(xlO ^)
in TiO^
2-x
Table 4.
P
(atm)
^2
Resistivity
(S^-cm)
14
1
10-3
2
10-1°
5.0
7
10-12
1.7
10
10-14
0.50
The Resistivity of Flux Grown Titania Oxides as a
Function of Large Values of X in TiO^_ at 27°C.20
X
in TiO^
2-x
Resistivity
(^-cm)
0.125
3.2 X 10-2
0.167
1.3 X 10-2
0.200
1.9 X IQ-^
0.250
6.3 X 10-4
0.333
7.1 X 10-1
23
tion of TiOp greatly decreases the resistivity from the
stoichiometric, room temperature value of 10 13 r^-cm.
Titanium dioxide experiences electrical aging when an
electric field is applied.
During this process one observes
evolution of oxygen from the specimen
reduces the resistivity.
which seemingly
An increase in the oxygen vacancy
concentration causes a change in the resistivity as a function
of temperature.
The electrical aging of titania is at least
one of the causes of the reduction of the spark plug resistance values during use in an automobile engine.
Properties of Boron Carbide
Boron carbide is a semiconductor that exhibits the
expected negative temperature coefficient of electrical
resistivity.
The literature contains many temperature-
resistivity studies of boron carbide.
Various results
have been reported and the results of Golikova, et.al., are
given in Figure 6.
Most of the literature reports that B^C does not
25
oxidize below 600°C, although Litz and Mercuir
report
slight oxidation as low as 450°C. Nazarchuk and
26
Mekhanoshiva,
using B,C in the size range of 62 to 74ym,
report that the oxidation increases at 700°C, it levels off
between 800-1000°C, and it increases sharply at 1200°C.
They
say the leveling off range is due to the formation of 62©^,
which protects the B.C from attack by oxygen.
The '^2^'^
volatilizes rapidly at 1200°C and the oxidation of the B^C
24
1700
700
-4
\
0.5
1
300
1.5
2
1/T K X 10
Figure 6.
0°C
100
2.5
3
3
3.5
The Electrical Resistivity of Boron Carbide 21
25
continues.
They also claim that B.C invariably contains a
minimal amount of carbon which reduces the oxidation resistance of the B.C.
Nazarchuk and Mekhanoshiva farther report that
coarse powders of B C form an outer vitreous layer at 800°C
in oxygen and do not burn up, whereas fine powders oxidize
as freely as carbon.
Litz and Mercuri use B.C in the size
range of 75 to lOSym and report that after their experiments
27
the B.C particles were covered with vitreous B2O0.
Dominey
concludes that a 626^ film reduces the oxidization rate.
He
also says that decreasing the particle size and the inherent
splitting of the B.C particles due to oxidization increase
the oxidization rate.
Properties of Borosilicate Glasses
Borosilicate glasses have a high electrical resistivity
relative to soda silicates and the temperature coefficient of
28
29
electrical resistivity is negative.
Westhuizen, et al.,
has shown that remelting Pyrex (13 w/o B^O-) does not signifi30
cantly alter the resistivity. But Dgebuadze and Mazurin
report that the resistivity of soda silicates with 26 w/o
and 37 w/o B^O-. increases, respectively, 2.2 and 0.7 orders of
magnitude after heat treatment.
Borosilicates are resistant to both heat and chemical
attack.
When heated at 1200°C for one hour, the volatiliza-
tion, measured by weight loss, is less than 0.1 weight per31
32
cent according to Oldfield and Wright.
Heslop
reports
26
that at 150°C there is negligible attack by most acids and
alkalines, hydrocarbons, and oxidizing or reducing chemicals.
Wilson and Carter 33 have shown that the diffusion
coefficient of sodium in Pyrex is less than that in a sodalime glass.
High melting point materials can be heated with
a borosilicate glass to its softening point and, after cooling,
the glass will serve as a rigid structure to hold or bond
together the materials.
Binder Effects
A binder facilitates the handling of powdered materials.
When a silicone resin binder is heat treated at a sufficient
temperature and time, it is completely transformed to silica
which must then be considered in the composition.
As the
methyl group is burned off, it combines with oxygen that is
both in the air and in any other components present which
are subject to reduction.
27
CHAPTER III
MATERIALS, PROCEDURES AND EQUIPMENT
This chapter describes the specific materials used
in developing the trial resistor mixes.
The materials were
titanium dioxide, boron carbide, a borosilicate glass and
a silicone resin binder.
A description is given of all
procedures and equipment for handling the materials, formulating the trial resistor mixes, sealing the mixes in
spark plugs, measuring the spark plug resistance values, and
sectioning the spark plug for microstructural examination.
Components of the Resistor Mix
The titania had the anatase structure and was a Baker
Analyzed Reagent, (lot no. 38356).
It was passed through a
100 mesh screen to break down conglomerates in order to promote uniform mixing.
The boron carbide was produced by
Cerac (IPB-25-13047-64; stock no. 1285), with 7.5 w/o between
-270 and +325 mesh, and the balance -325 mesh.
The binder
was Dow Corning Release Coating, a silicone resin binder
originally developed for use in bakery bread pans.
The low
viscosity of this binder gave it good wetting characteristics.
The composition is proprietary but the binder oxidizes to silica
when heat treated in air.
is given in Appendix A.
Further heat treating information
28
The borosilicate glass was Corning 7070 and is
described by Hutchms and Harrington.
34
Corning 7070 is a
low loss electrical type glass with the following composi71 SiO^, 26 B^O^,
tion (w/o):
1 Al20^, 1 K^O, 0.5 Li20,
and 0.5 Na20. The thermal expansion from 0°C to 300°C is
—6
3.2 X 10 /°C, and from room temperature to 460°C it is 3.9
X lO" °C.
CDS.
The strain point is 455°C at a viscosity of 10
Figure 7 shows the volume resistivity as a function of
temperature.
Three specific values of the log,-, volume resis-
tivity iQ,-cm)
given by Hutchins and Harrington at 25, 250,
and 350°C are, respectively, 17+ (estimated), 11.2 and 9.1.
Mixing the Components and Loading the Insulators
The TiOp and the B^C were weighed using an Ainsworth
analytical balance and the measurements were accurate to
+0.01 g.
The glass and the binder were weighed using an
Ohaus triple beam balance and the measurements were accurate
to +0.5 g.
In each series of trial resistor mixes, the
effect on the spark plug resistance values of the amount of
one component was studied.
A large amount of the other
components (except the binder) was mixed and then divided
into equal amounts.
The component being studied, either the
TiOp, B,C or binder, was added in the various amounts.
all cases the binder was added last.
mix weighed 25 to 30 g total.
In
Each trial resistor
29
13
e
o
I
B
•^
12
>i
-p
•H
>
m
•H
m
(U
oi
^
10
rH
O
>
o
rH
D^
O
9 -t
H^
200
300
400
Temperature
Figure 7.
500
o.
(^C)
The Volume Resistivity of Corning 7070 Glass 34
30
The dry components were mixed by hand until the TiO^
and B.C appeared well dispersed in the glass.
The binder
was added and hand mixing was continued for about five minutes
until the binder was absorbed by the material and wetting
appeared uniform.
The mixes were dried in air and carefully
granulated with a mortar and pestle.
They were finally
screened to yield 95 w/o of -28, +100 mesh and 5 w/o of
-100 mesh.
Spark plug insulators, terminal screws, center wires
and a copper glass were obtained from a spark plug manufacturer.
Descriptions of these are given in Appendix B.
The
copper glass was a layer on top of the resistor mix in the
insulator bore to assure good electrical contact with the
terminal screw.
The insulator was loaded by volume to yield
193 +5 mg of resistor mix and 124 +5 mg of copper glass.
The volume loader was a brass bar, one each for the resistor
mix and the copper glass, with a hole bored in it.
The
material was poured into the hole and the excess was scraped
off using a spatula, the edge of which was level on the brass
bar.
The resistor mix v/as poured into the insulator on the
head of the center wire, the copper glass was poured on top
of the resistor mix, and the terminal screw was placed on
top, extending above the insulator as shov/n in Figure 8.
This assembly was sealed as described in the next section.
Sealing Apparatus and Procedure
The sealing apparatus was built to individually seal
31
>r~---y
/
\ I
\ I
Figure 8.
A Schematic Drawing of Sealing a
Loaded Resistor Spark Plug.
32
spark plugs in a manner similar to a large scale process.
The times and temperatures of the sealing procedure were
selected because they simulated a typical production process.
Briefly, the spark plugs were heated for 5 min. 4 0 sec.,
cooled for 10 sec. and hot pressed for 28 sec.
They were
cooled in air to room temperature.
The sealing apparatus is shown in Figure 9.
The spark
plug sat on a stool (hollow steel cylinder) with three gasair torches equally spaced around it.
Each flame was at an
angle of 50+5° from the horizontal and impinged the stool at
1 cm below the top.
Figure 8 shows the spark plug on the
stool being heated.
For simplicity only one torch is shov/n.
The temperature was controlled by the pressure of the
gas and air going to the torches and the control knobs on the
torches were fixed.
The air pressure was 12 psi, measured
on a 0 to 60 psi pressure guage.
The gas pressure was mea-
sured by the number of turns of the gas valve which had a
very reproducible and linear ratio of number of turns to
pressure.
A monometer v/as later installed and used, and the
gas pressure was 13 inches of water.
The pressures given above were used because they produced a flame which made the interior of the insulator 9 75
+15°C.
This temperature was periodically measured using a
chrome1-alumel thermocouple.
The bead of the thermocouple
was embedded in the resistor mix of an insulator that had
only the center wire and the resistor mix in it.
This
33
(A)
(B)
,.--n'>';iSee3^;
Figure 9.
The Apparatus Built to Seal Spark Plugs in
a Manner Typical of Production Techniques:
(A) General View, (B) Close-Up of the Stool
and Torches.
34
assembly was heated in the sealing apparatus and the temperature was monitored from room temperature to the end of the
heating cycle, 5 min. , 4 0 sec.
Maximura temperature was
achieved between three and four minutes after initial heating.
There was a metal plate located 6 cm above the terminal screw.
It weighed 14 kg and could be raised and lowered.
It was used to press the spark plugs before heating, which
compacted the resistor mix and the copper-glass.
This
increased the density of the final sealed resistor mix by
allowing more pressure to be transmitted through the molten
materials during hot pressing.
The same plate was used for
hot pressing the terminal screw into the insulator.
Three procedures were used for sealing the spark plugs.
In all three the spark plugs were pre-pressed on a cool stool.
For procedure A, the spark plugs were heated for 5 min., 40
s e c , from room temperature to 975 +15°C.
They were cooled
for 10 sec. during which time they were transferred to a
cool stool.
The plugs were hot pressed for 2 8 sec., removed
from the stool, and allowed to cool in air to room temperature.
Procedure B was similar to Procedure A, the only difference
was that the hot plug was not transferred to a cool stool for
hot pressing.
Procedure C allowed continuous monitoring of the spark
plug resistance values during the sealing procedure.
Before
loading,chromel wires were spot welded to the electrodes.
These leads were connected to a millivolt recorder with special
35
circuitry and calibration for monitoring resistance.
The
circuitry is shown in Figure 10, and a typical resistance
plot is shown in Figure 11.
Otherwise, Procedure C was the
same as Procedure B.
Resistance Values
The millivolt recorder was calibrated from 100 to 10
mV as 0 to 100 kn,
box.
respectively, using a standard resistance
Resistance values greater than 100 kO, (from 0 to 10 mV)
were not measured, but designated as >100 kQ.
The sealed
plugs were allowed to cool at least three hours before the
room temperature resistance values (Rr^m) were measured using
R.-L
the millivolt recorder.
The mean, X, and the standard
deviation, a, were calculated for the final resistance values
of each series of plugs containing the same resistor mix.
For the standard deviation,
o 1/2
a = [-Kr
Z (X.-X) ]
^-1 i = 1 ^
(1)
When the millivolt recorder was used (Procedure C ) , the minimum resistance value was also recorded since it varied with
some compositional changes.
This sealing temperature value
(R„ ) occurred just before hot pressing.
Attempts were made to reduce the scatter in the resistance values by increasing the B.C content, reducing the
particle sizes of the components, and changing the geometry
of the electrodes.
Limited success was achieved with the last
36
4.5 V
100 Q.
to
spark
plug
Electrodes
to
10 mv
recorder
A/W10 kQ
Figure 10.
The Circuitry Used to Modify the Millivolt
Recorder to Measure Resistance Values of the
o cr
Spark Plug Throughout the Sealing Process
.
37
M
rH
u
fd
•
iH
Ui
VD
m
m
<D
Q)
u
u
:3 0
W i-l
•
CN
td a.
(U
S tP
If)
+J rH
cd
CJ
0 -H
•
'^
^^^
G
M
i-H
•
r-
in
o
•
rH
*—'
Q)
U
C
td
-P
to
•H
CQ
Q)
P^
Cn
:3
<^
r-i
iH
CM
44
>-J
fd
LO
CN
• ^
• ^
ro
o
rH
o
o
o
o
o
r^
o
o
V£>
o
o
in
{DBS)
o
o
• ^
BUITJ,
o
o
ro
o
o
O
CN
r-i
bUTJ-QBS
O
o
Pu
m
Tl 0
Q) W
-P
fd
U
JQ
•H
rH
<u
Xi
-P
(T>
fd c
U -H
M
-P
0
3
Q
iH
Oi W
<U
(U
T!
U
0
O
Q)
Cti
H
fd
>
QJ
O
C
td
-P
cn
-H
tfi
(U
P::
-P
iH
0
>
•H
r-l
rH
•H tT>
S 13
rH
<: a^
0
U
13
Cn
•H
38
method, and the information is given in Appendix C.
Photomicrographs
In order to study the microstructure of the sealed
spark plugs, the plugs were sectioned axially, ground and
polished, and photographed.
Figure 12 is a typical photo-
micrograph showing both electrodes, the copper glass and
the resistor mix.
39
Figure 12.
A Picture Showing the Resistor Component,
Copper-Glass and Electrodes Sealed in an
Insulator.
40
CHAPTER IV
RESULTS AND DISCUSSION
Introduction
Trial resistor mix compositions were formulated using
some or all of the four components to study the effect of
each component on resistance values.
Initially, the Ti02^
B.C and binder (hereafter called modifiers) were added to
the glass individually and in pairs.
Subsequently, all
three modifiers were added to the glass in various proportions
The specific resistor mix compositions, the resulting spark
plug resistance values, and a discussion of these results
are given in this chapter.
Based on the discussion of results
and a microstructural examination of the resistor component,
a model is postulated for the current flow through the resistor component.
Resistor Mix ID:
N, 9, and Q
Titania, B.C and binder were added individually to
the glass for mixes, N, 9, and Q, respectively.
The two com-
ponent compositions and the resulting resistances are shown
in Table 5 and Figure 13.
(In Tables 5, 6, and 7, R^^^ is
the mean spark plug resistance value at room temperature,
R
is the mean value, if measured, at the sealing tempera-
ture, a,
is the standard deviation of the mean, and n is the
41
Table 5.
The Compositions and Resulting Spark Plug
Resistances for Two Component Resistor
Mixes: N, 9, and Q.
Weight Percents
Average (kf^)
^ST
^RT
^
93
TiO^
B-C
Bind(
2
4
Glass and TiO^
7
-
7.9
>100
2
N2
89
11
6.6
>100
2
N3
81. 3
18.7
0.4
>100
2
N4
87.,8
12.2
>100
3
3.3
>100
2
5.0
>100
2
2.6
>100
2
Mix ID
Glass
Nl
-
-
Glass and B^C
el
99
-
1-0
02
98.5
-
1.5
83
97
-
3
-
Glass and Binder
Ql
95
Q3
85
_
_
5
2.7
>100
2
^
_
15
0.5
>100
2
42
A w/o Ti02
7
O w/o B^C
D w/o Binder
5
Figure 13.
8
9 10 11 12 13 14 15 16 17 18 19
1
2
6
7
3
8
9 10 11 12 13 14 15 16 17
The Minimum Resistance Values Measured
During the Heating of Two Component
Resistor Mixes: N, 0 and Q.
43
number of spark plugs that were sealed v/ith each trial
resistor mix.)
None of the components added individually
to glass lowered the room temperature resistance below 100 kQ,
Increased additions of TiO^ and binder each continually
lowered the sealing temperature resistance, but increased
additions of B,C did not show this trend.
Resistor Mix ID:
P, R, and S
These mixes had two modifiers added to the glass,
TiOp and B.C, TiO^ and binder, and B^C and binder, for mixes
P, R, and S, respectively.
The three component compositions
and the resulting resistances are shown in Table 6 and Figure
14.
A review of the data in Table 6 shows that TiO^ must be
in the resistor mix to get the resistance values below 100 kJ2.
It was tentatively concluded that TiO^ is the primary conducting medium in the resistor mix, although the B.C and
binder contents do affect the room temperature resistance.
Note that some of the data for the P series appear twice in
Table 6 to show different trends, i.e., one modifier or ratio
of modifiers was held constant while the other increased.
The data is similarly arranged for the R series.
Resistor Mix ID:
T, U, and V
These mixes contained all three modifiers; in the T,
U, and V series, respectively, the binder content, the TiO^
content, and the B.C content were varied.
The specific com-
positions and the resulting resistance values are shown in
44
'd
<U
:3
•H
4->
(U
Q)
U
O
U
^
LO
(O
ro
ro
CN
rH
CN
(N
rH
iH
O
•
r-i
o
c:
ro
CM
CM
0
LO
O
IT)
m
u
o
M-l
m
Q)
V£)
o
^
•
(13
4-»
CO
•H
CJ
(U
-p
p:^
d^
V£>
CN
•
ci4 cn
^
T(
(TJ fU
Oi
tn
0)
1T>
E-t
H
(^
o
-P
:3
CO W
Q) Q)
Pi X
•H
M
O
en -P
fd
o
W
•H
cn
-P 0)
c:;
m -P
o
e c
o OJ
u o
•H
Q)
x: uo
EH
(JD
Q)
rH
Xi
(d
EH
cn
cn
•^
CN
•
O
ro
r-\
C3
rH
,
•
O
o
fd
0
CN
• H p4
iH
U
Q)
TH
G
•H
PQ
•H
EH
}^
fd
H
EH
1
tn
tn
fd
U
U
1
"^
1
1
CQ
PQ
nH
iH
CM
iH
iH
r-i
1—<
rH
rH
H
•
Q)
fl^
CM
+J
O
•
m
tp
c:
•H
^
fd
CN
O
•H
EH
cu
u
o
•
VO
CM
LO
•^^f
O
rH
CN
cn
fd
t
O
o
•
H
CM
1
1
cn
cn
fd
PQ
Cn
•H
cn
fd
Q)
O
u
•
CD
rH
cn
o
in
U
o
O
•H - ffi
c:
Cn
^
cn
O
rH
iH
0)
o
-^ •H
CQ
EH
o
CM • H
U
+J
•H
0)
u
O
Gn
nj
5-1
Q)
>
<
•<*
u"d^
w ^
en
•H
•H
+J
tn
0)
iH
CN
viJ
iH
o
1
1
•H
EH
iH
O
tH
ro
rH
H
fi
O^
<y\
r-i
iH
rH
c^
CO
CO
00
u
u
f-\
rH
O
•H
EH
•^
cn
cn
td
I—)
o
in
CO
00
r--
r^
o
00
00
00
r-i
Oi
LO
CM
ft,
a.
ro
o
00
Q
H
•H
fl^
"^
CM
C^
ro
P4
45
en
en
m
m
r-i
m
ro
CN
rsj
ro
O
O
o
o
o
o
o
-p
in
in
o
o
m
Q)
+J
M
O
O
EH
D:;
Q)
>
j::
H
03
fd
W
P:^
J
oCN
Q)
T5
c
0)
o
-H
CQ
o
LO
(D
CN
CN
•H
EH
03
W
Id
U
o
-H
O
in
EH
CO
CN
on
in
ro
CN
CM
Ti
m
•H
CQ
U3
03
fd
r-\
u
•H
CQ
pq
in
O
U
"^
cq
C^
c;
•H
03
fd
CO
en
ro
VD
CM
VJD
0
u
O
CJ
H
CM
CT^
in
(7^
in
CT>
CO
CN
r-^
in
00
CN
CN
r-
CM
CO
cr»
00
ro
O
in
tr>
5H
Q)
-
'O
03
03
fd
iH
C
-H
CQ
O
•H
03
fd
0)
}H
CQ
•H
fd
I
rH
A
U
G)
'd
Q)
">^
o
•«*
o
o
»:>
CO
en
fd
rH
vo
Q
H
ON
H
00
<T\
"vT
CO
o
CO
rH
rH
EH
C
O
U
td
EH
m
Q)
H
U
o
©
rH
U
Q)
rH
-p
X3
Cn • H
+)
-H
CN
CNJ
•H
•H
CN
•H
ei4
•H
(D
U
Q)
TJ
o
o
fd
0)
CD
13
CN
Ti
: u
J
to
-p
cn
tJ>
W
(1)
Cn
fd
U
o
u
CN
CM
in
a\
in
in
CM
iH
CO
o
CN
X
•H
p^
ro
P^
P^
in
P^
UD
P^
W
00
CM
ro
C/3
W
46
[3
a,
^ 24
Cfi
(D
^
rH
(Tl
22
11 w/o TiO„
PI, P2 & P5
20
>
o; 18
u
n3 16
-P
en
•H
w
14
0)
12
'E
10
B
0)
&H
e
0
O
K
8
b
^B
19 w/o Ti02
P3 & P4
4
rH
(G
C
2
•H
fc
n
1.0
2.0
3.0
w/o B C
Figure 14.
The Final Resistc3nce Values as a Function of
w/o B;,C at Two TiO Contents in a Resistor
4
2
Mix Composed of Glass, TiO^ and B.C.
47
Table 7 and Figure 15.
In the T series, the increase in resistance values
as binder was added up to about 6 w/o was not anticipated.
Adding binder was expected to reduce the TiO^, making it
more conductive and reducing the overall resistance of the
spark plug.
This unexpected behavior could possibly be
accounted for as follows.
When the binder burned, it combined with oxygen to
form Si02/ CO2 and water vapor (see Appendix A ) .
The binder
first consumed the oxygen in the air in the resistor mix and
then attacked the Ti02 ^^-^ additional oxygen.
At low binder
content, the oxygen in the air was sufficient to satisfy the
binder so that no Ti02 was attacked.
As it burned the binder
left behind small crevices around the resistor mix particles
due to the physical absence of some binder molecules.
silica content was also increased.
The
These crevices and
additional silica served to increase the resistance value of
the resistor mix from 0 to about 6 w/o.
Around 6 w/o, there
was enough binder to attack the Ti02 for oxygen.
The Ti02
v/as reduced and became more (relatively) conductive as the
binder content was further increased.
As more Ti02 was
reduced, its conductive effect ovei'came the resistive effect
of the crevices and the additional silica, and the resistance
value of the mix decreased.
Thus, a higher binder content
led to low resistance values.
The effect of the binder content between 9 and 15 w/o
48
M
13
0
0)
•
&4
c:
M
O
LD
LO
vo
^
LO
in
00
CO
00
m
O
U
0)
o
G
O
03
-M
to
b
iH
_
a
^
—'
en
3
iH
p^
en
Q)
Cn
n3
ri<i
J^
fO
a.
w
LO
o
CN
rH
o
in
•^
•
CO
<-^
r-
rH
o
r-^
•
EH
p:;
Q^
c >
(I>
>
<
EH
CO
(i;
0
^
cr; D
-
CO
u
^
(U
04
•H S
to
0 -P
-p
0
U
G
0
0^
0) g
x; 0
O
LO
CN
r-
x:
en
•H
VJD
ii
•
"^r
rH
r-
(0
r^
m
•
CO
fd
r-{
a
CN
o
•H
pq
^
tn
C
U
"^
(Q
O
CN
<u
c
•H
CQ
m
-M
C
(U
4-> " H
a G
s <u
o
CO
CO
(U
Ti
5H
CO
0 (U
•H X
m
rH
rH
O
o
II
r^
o
00
VO
rH
CO
r-\
CO
r-^
LO
o
ro
O
\
o
o
o
O
O
o
u^
m
to
«> to
U
(T3
-P TJ
.H G
^ fO
to ••
•
CO
rH
•H
EH
•H
EH
VO
rH
\
•H
CO
(d
(U
^
o
G
o
•
o
rH
O
•
ro
V£)
•
r-\
O
•
as
CN
•
"sf
o
•
LO
r-i
CQ
O
•H
EH
•
o
LO
rH
-•^
<D
CM
^
ro
ro
00
CNJ
rH
rH
r-^
T-^
rH
r-i
SH
o
c:
H
CO
fd
0)
rH
CN
vo
LO
CN
CO
CO
CN
•
•
o
o
r^
^
<-\
•
LO
CO
00
(d
\
•<:r
JH
•
0)
Id
<T>
<y\
O
VO
rH
^
ro
•
ro
•
cH
f-i
iH
CN
O
o
t-{
•
00
LO
ro
o
vo
ro
•H
EH
iH
rH
o
o
<j\
C3^
f-\
O
rH
to
to
CO
CN
LO
o
•"^
iH
r^
-^
CN
CO
cr\
r^
rr-
^
r-
CN
00
r-{
CN
EH
ro
"^r
r-i
EH
LO
EH
*^
EH
O
r-i
o
r^
H
as
•
•
II
Cn c:
CJ •H
•H
PQ
LO
•^
•
CN
LO
rH
rH
s. t-i
CN O
rH
O
\
U
ro
o
CN
•
rH
O
II
rH
r-i
•
^
rH
r^
O
JH
ir- •
EH
'sf
o
CN
rH
0)
ff;
C
rH
r-m
•H
CQ
TJ
C
fO
•
r^
rH
•
•
o
•
m
ro
«
cr»
r--
r•
vo
r-
rvj
D
ro
D
T-i
r-i
0
IS
CO
ftJ
iH
CO
00
•
O
r(U
•H
-P
Q
iH
H
^
(^
X
EH
•H
^
EH
D
49
VX)
KD
KO
v^
LO
^
o
U3
VO
^
LO
CN
O
'=?•
vr>
o
LO
ro
o
"«^
CN
(N
CN
rH
m
00
O
<y>
ro
LO
0
ro
0
0
LO
0
ro
"^r
r-\
0
LO
cr>
o^
0
<y\
00
LO
«0
ro
ro
CN
H
rH
0
0
0
1
i
1
1
rH
H
a\
vo
0
rH
0
rH
0
rH
03
0
0
0
CN
to
(D
rH
r-^
C5
^
EH
C^
0
II
01
a)
03
ifj
CP
TCf
j-j
tH
(U
(T3
rH
>
\.
vD
CN
ro
rH
0
1
1
!
0
10
cr»
a>
CO
0
r-\
0
>-i
0
r-i
CT>
<y\
cr>
cr\
ro
<y\
LO
0
CO
0
0
0
•^
00
t--
0
CN
iH
rH
n-i
r-\
CN
CN
CN
ro
<y\
<y\
r~
c:o
CO
r-
r-
^D
KD
0
rH
0
rH
0
rH
0
E-i
0
r-A
0
r-\
0
f-i
0
rH
0
rH
Cfi
0
CO
cr>
rH
'^
CO
ON
CM
ro
rd
00
r-
VD
i^
<o
r-
r-
KO
f"~
KD
(^
r-
UD
[^
rr-
ro
"^
>
LO
^o
>
r>
CO
cr\
0
cn
21
1
CN 03
03
-H
03
r
H
EH
r-i
0
^
Q)
0
•H
u
CQ
CQ
T3
C
•H
CQ
c
tn
j::
0
u
u
<u
U
CQ
fN
4->
O
•H
0)
c
IS
•H
4J
fi
0
o
•H
0
fll
T3
0)
:3
U
"^ 0)
'd
1
tn •H
w
iH
O
H
r-
<-^
v^
•
Q
<D
H
XJ
fO
X
•H
2:
rH
EH
CN
>
>
>
>
>
0
r-i
>
50
-p
!
•
C u
0
,,-v
Q) <U
-M TJ
use
C 0 -H
fO U PQ
-P
m x ; TJ
•H +J C
W -H fd
5i ^
rt > 1 u^
en M cq
:3 fO
rH >
-
CN
w O
,i4 0) -H
S-l :3 EH
(d rH
CU
m
0
en
-P
C
0)
c;
0
a
e
0
U
a
M
--
W
Q) 0)
^ P
:3 M
-p fc
c
(U
fO >
}^
u
0 Q)
u
(U U) Hi O
Ul
-p
&< w e e
(T3 0 (0
-P iH EH -P
m
^ O
Cn 1 g -H
•H c o w
51
is not certain.
The 15 w/o data point is probably in error
since this resistance value has a very high standard deviation.
Approximately 10 w/o binder was used in all subsequent
resistor mixes.
This value was enough to insure uniform
mixing and above the 6 w/o addition that gave the high resistance values.
In order to determine qualitatively if any binder was
not completely burned during the sealing process, five plugs
were reheated in the sealing apparatus and visual estimates
were made of the amount that the terminal screw rose during
reheating.
Three of the plugs were from the P2, P3, and P5
lots which contained no binder, and they rose 1/8, 1/16 and
1/8 inch, respectively.
This behavior was due to the expan-
sion of trapped gases as v/ell as the expansion of the hot
materials.
Two of the plugs were from lot T4
which con-
tained 9.1 w/o binder, and they rose 3/16 and 1/4 inch.
Since
the plugs with the binder rose more during reheating than the
plugs without binder, it is assumed that there was residual
binder in the resistor mix after the original heating.
The
effect of the unburned binder on the resistance values is
unknown.
In the U series of trial resistor mixes, the effect of
the TiOp content was studied.
The resistance values decreased
from 4.6 to 1.9 as the TiO^ was increased from 7 to 13 vz/o.
This behavior was expected because of the increasing amount
of reduceable TiO^.
A comparison of mixes U2 and T4 shcv/s
52
that the resistance values in the U series were somewhat low,
but this was probably due to slighly different sealing conditions as discussed in the procedure.
In Table 3 of Chapter II, the large increase in the
conductivity of TiO^, even at slight reduction, was shown.
The S series (Table 6) contained no Ti02 and all the mixes
yielded resistance values >100 kfi.
In Table 7 the amount of
Ti02 is critical below 7 w/o, based on a qualitative comparison of the S series and the U series and on the information
referred to in Table 3.
The conclusion from this information
is that Ti02 is considered the primary conductor in the resistor mix, with the B.C and the binder having secondary effects.
Considering that a slight reduction of TiO^ greatly
decreases its resistivity (see Table 3 ) , and considering the
decrease in the resistance values due to the addition of Ti02
(see the S and U series. Tables 6 and 7 ) , it is concluded
that the Ti02 is reduced during the sealing procedure.
The
extent of the reduction is assumed to be slight, but the
amount is unknown.
In the V series, the B.C content was not studied
below 1 w/o because there may be a problem of effectively dispersing the B.C particles.
The effect of changing
the B.C
content on the resistance values was greatest around 1 w/o
and resistance values were almost independent of 3.C content
above 1.8 vz/o.
The B,C curve in Figure 15 agreed well with
expectations based on the P series.
According to Figure 6,
53
the resistivity of B.C is about l„Q-cm at room temperature
and so this component adds significantly to the conduction
of the resistor mix.
Microstructural Examination of the Resistor
Component and the Copper Glass
This discussion is an analysis of the microstructure
of the resistor component and the copper glass of a sealed
spark plug.
Typical photomicrographs, shown in Figures 16,
17, and 18, reveal a laminar geometry of the resistor component.
The photomicrographs are oriented with the terminal
screw above (at the top of the page) and the center wire
below (at the bottom).
In each figure, part (a) is a light
field view and part (b) is a dark field viexv, which allows
subsurface observation.
The photomicrographs of the microstructure revealed
that a laminar flow of the resistor mix occurred during hot
pressing when the material was extruded up around the terminal
screw and down around the center wire.
Between the electrodes,
the flow was approximately parallel to the center wire head,
and on the outer edges, the flov7 was parallel to the insulator
walls, as shown in Figure 12.
The laminar flow consisted of
alternate layers of glass and TiO^, with B.C particles and
pores randomly oriented in the flow matrix.
The dark field views particularly reveal the laminar
pattern.
During the heating period, the glass particles
softened enough to be deformed and to flow during hot-pressing.
54
(a)
»^
w^
•
^^>
•
•
'
•
-
^
•
»
-
-
^
*••* « ^ i ^ J ^ ' 1 * _
jr--»
Figure 16.
*
'ijs^•'
Photomicrograph of a Sealed Resistor Component
Revealing that Lam.inar Flow of the Material
Occurred During Hot Pressing: (A) TiOn Layers
(B) Glass Layers, (C) B^C Particles (542x).
55
(a)
(b)
Figure 17.
Photomicrograph of a Sealed Spark Plug Showing
the Copper-Glass and the Resistor Component:
(A) Copper-Glass, (B) Ti02 Layers, (C) B4C
Particle, (D) Pores in the Glass-Ti02 Matrix,
(E) C o p p e r - G l a s s Belov; t h e S u r f a c e of t h e S a m p l e ,
(F) Ti02 in the Pores
(196x).
56
Figure 18.
A Photomicrograph Showing the Interface Between
the Sealed Resistor Component and the Center
Wire Head: (A) Center Wire Head, (B) B.C
Particles, (C) Pores in the Glass-TiO^ Matrix,
(D) Ti02 in the Pores (196x).
57
Heating was insufficient for the glass particles to flow to
the extent of removing all grain boundaries.
In Figure 16,
the Ti02 is seen in alternate layers with the glass.
The
TiO^ layers are in the glass grain boundaries but are not
continuous because the glass separates some of the individual
particles and conglomerates.
All three figures show pores and B .C particles randomly
orientated in the glass matrix.
In some cases, the pores
are like crevices which are deformed parallel to the glass.
Some Ti02 can be seen in the pores at 542x (Figure 16), and
when the pores were examined at 945x, TiO^ particles were seen
as lining on the surfaces of the pores.
There were several
possible causes of the formation of the pores.
Pieces of
glass may have been "pulled out" of the sample during grinding.
The pores may have been pockets of trapped gases created by
the decomposition of the binder or air that became trapped as
the glass softened.
The interface between the copper glass and the resistor mix as shown in Figure 17, had no distinct boundary.
The
dark field view shov/s the copper particles below the surface,
and the copper appears to make a continuous path from the
resistor mix to the terminal screv/.
The interface between the center wire head and the
resistor mix, as displayed in Figure 18, shows some type of
layered material on the surface of the center wire head, which
may be an oxidation layer.
Otherwise, there appears little.
58
if any, reaction among the components.
At least eight B C particles can be seen in the photomicrographs in Figure 18.
Particularly the one in the upper
left corner and the two in the lov/er middle section are in
positions associated v/ith several layers of TiO^.
Note the
overlap of the lower right part and upper left part of
Figures 17 and 18, respectively.
Model for Current Flow Through the Resistor Mix
Based on the discussion of the compositional effects on
resistance and on the microstructureal examination, a model is
postulated to explain the role of each component in the conduction of the electric current through the resistor component.
Several further conditions for the model follow.
Measurements for the dimensions of the components after being
sealed in the insulator were estimated from the photomicrographs.
The Ti02 layers are 0.5 to 3 ym thick and are assumed
to be electrically continuous.
There are some agglomerates of
Ti02 that have larger dim.ensions.
The B.C particles are 10
to 30 ym in size and the glass layers are 10 to 50 ym thick.
Before sealing, the resistor mix powder consisted of agglomerated fragments of glass, each generally surrounded by a thin
layer of 1'i02^ with a few B.C particles dispersed among the
glass pieces.
In Appendix D, a calculation shows that there
is enough TiO^ in the resistor mix to create the number of
Ti02 layers shown in the typical photomicrographs.
In the model, current flows through the terminal screw
59
and the copper glass with negligible resistance, and it can
flow from the copper glass to the resistor mix at any mutual
contact point.
Each layer of glass in the resistor mix is
an insulator and each layer of TiO^ acts as a conducting path.
The pores are resistors but the TiO^ particles lining the pore
surfaces act as conductors.
ductors.
The B.C particles are also con-
The resistor matrix is then a circuit of conductors
in series and in parallel.
Further, a piece of glass deforms horizontally betv/een
the electrodes during the hot-pressing, and the fine TiO^
particles form broad layers in the glass grain boundaries.
The current flows on these Ti02 layers and through the B.C
particles which are in the grain boundaries in the glass
matrix.
The coarse B.C particles are too hard to deform, but
the size of the particles allows the "connecting" of several
layers of Ti02.
'^^is "connecting" behavior occurs where the
B.C particle size is similar to the thickness of the glass
layers, as shown by the dimensions gi\^en above.
The TiO^ is
the primary conducting medium because it flows and spreads
into layers with the glass.
The B.C is also significant
because of its "connecting" function.
The laminar matrix
that forms parallel to the insulator walls is similar to the
matrix between the electrodes except for orientation.
A simple schematic of the resistor mix betv/een the
electrodes and a similar model of the resistor mix next to the
walls of the insulator bore are shown in Figure 19. The glass
60
B C Particles
4
Pores with TiO.
Copper-Glass Extremities
Center Wire Head
(A)
Fiqure 19.
(B)
A Schematic Drawing of the Resistor Component
Showing Possible Current Paths Through (A) The
Area Between the Electrodes and (B) The Area
Alonq the Edges of the Insulator Bore.
61
is represented by the clear, unmarked background.
are the layers of conducting TiO^.
The lines
For simplicity they are
drawn continuously indicating that the horizontal layers of
Ti02 are assumed to be electrically continuous.
particles and the pores are shown as labeled.
The B.C
The current
flows from the copper glass to the center wire head through
the shortest paths.
It flows along the Ti02 layers, through
the B.C particles and around the pores on the Ti02 that lines
the pores.
The role of the binder in the current flow is not known
at this time.
Evidence was shown which suggests that some
unburned binder remains in the resistor mix after the initial
heating.
Also, the burned binder produces some silica and
carbon, and possibly some pores.
Because of these uncer-
tainties, no attempt is made in this model to define the role
of the binder in the conduction of the electrical current
through the resistor.
The model does assume that the binder
serves to reduce the Ti02 during the heating cycle.
This
conclusion is based on a comparison of the N and R mixes
which shows that the addition of the binder to glass and TiO^
reduced the resistance values from infinity to as low as
about 5 kQ,
Discussion of the Model
This discussion shows how the proposed model agrees
with the data in Tables 5, 6, and 7.
In the N mixes (Table 5 ) ,
62
as much as 18.7 w/o TiO^ was added to the glass with no
resulting resistance values below 100 kS7.
(Table 6 ) , 11 w/o TiO^ and 1 w/o B.C
values to 2 6 kn.
In the P mixes
lowered the resistance
This behavior shows the "connecting" effect
of the B,C as proposed in the model.
In the R mixes (Table .
6 ) , 10 w/o TiOp and 10 w/o binder lowered the resistance
values to 36 kfi.
Thus the binder reduced the Ti02 to the
extent that the "connecting" function of the B.C was not
necessarily needed to lower the resistance values below
100 kfi.
The data provided by the R and S mixes (Table 6)
showed that, even without any B.C, the TiO^ and binder contents
can be increased enough to lower the resistance values to
about 5 kf^.
\
I
But if the Ti02 is completely removed, the resis-
tance values of the samples containing B.C and binder will be
>100 kfi (for the compositions tested in the S mixes).
In the
proposed model, at a high TiO^ content, there would be enough
coiiducting layers for the current to flow to the center wire
without any B.C particles.
If only B.C v^ere present, the
particles would be isolated by the glass and the resistance
would be very high.
In a mix with all three modifiers, less
than 2 vz/o 3 .C is effective in reducing the resistance values
because the B.C connects several layers of conducting TiO^.
The gap between the copper-glass and center wire head
is short compared to the distance along the insulator bore
from the side of the copper-glass to the side of the center
63
wire.
But, in the gap, the laminar flow is perpendicular
to the shortest current path distance, and so the current
must "zigzag" on the conduction paths around the glassygrains.
On the contrary, the lam.inar flow pattern adjacent to
the insulator walls is parallel to the shortest direction of
current path, as shown in Figure 19.
So, the current can
flow relatively straight to its destination.
Information
in Appendix C shows that there is no correlation between
spark plug resistance values and the gap distance betv/een
the copper glass and the center vzire head.
According to this
model, "all" of the volume of the resistor material contributes to the measured resistance value and, hence, the total
resistance is relatively insensitive to dimensional variations in the bore of the insulator and m.etal electrodes.
64
CHAPTER V
CONCLUSIONS
In this study the compositional effects on the resistance value were investigated for a prototpye resistor mix
consisting of approximately 80 w/o of a borosilicate glass
with additions (called modifiers) of titania, boron carbide
and binder.
During the heating and sealing the hot extrusion
of the mix resulted in a laminated structure of the resistor mix components.
The influences of composition on resis-
tance are listed initially and then combined with the directional macroscopic nature of the material to provide a model
for the bulk resistance behavior.
1.
When the modifiers were added individually to the borosilicate glass, none of the final resistance values could
be lowered below 100 kfi within the ranges of compositions
studied.
2.
When the modifiers were added in pairs, titania had to be
one of the components in order to produce final resistance
values below 30 kf2.
3.
In compositions containing no binder the introduction of
B,C particles was very effective in reducing the final
spark plug resistance to values below 10 kfi.
4.
In resistor mixes containing no B.C particles more than
65
15 w/o each of Ti02 and binder were necessary to achieve
resistance values of 5 kc.
5.
The TiOp is the primary conductor and the B.C is the
secondary conductor in the resistor component.
6.
In compositions containing all components, the additions
greater than 5 w/o of the silicone resin base binder
appeared effective in lowering the resistance value.
This
behavior undoubtedly occurred because of the reduction of
the titania, which made this component more conductive.
Based on the microstructure examination of the prototype resistance after sealing the following conclusions
were made.
7.
During the sealing process the resistor mix was extruded
from between the electrodes up around the top terminal
screw and down around the. head of the center wire.
8.
During extrusion a very laminar structure was formed
consisting of elongated glass grains coated with the
titania powder.
The glass was oriented perpendicular to
the electrodes between the electrodes and parallel to the
electrodes along the bore wall of the insulator.
9.
The conduction paths for the current through the resistor
mix from the terminal screw to the center wire head consist of electrically continuous layers of TiO^ alternating
with layers of glass.
The B.C is dispersed randomly in
the resistor mix acting as connecting points betv^een the
layers of TiO^.
66
10.
The B,C particles served to prcvioe conducting links
between the layers of titania powder and, hence,
increasing the number of particles was effective in
reducing the resistance values.
11.
The anisotropy of the resistor macrostructure effectively
served to miniraize variable resistance values caused
by non-uniform dimensions on the various components of
the spark plugs.
For example, the information in
Appendix B shows no correlation of the gap distance
between the copper glass and center wire and the measured spark plug resistance values.
67
APPENDIX A
HEAT TREATMENT INFORMATION DOW CORTTING
RELEASE COATING Ql-2531
The binder used in the resistor mixes that were formulated in this investigation was Dow Corning Release Coating
Ql-2531 (lot # BH026036).
The material is a silicone resin
binder and the composition is proprietary.
When heated from
room temperature to 800°C for three hours and then held at
800°C for one hour, the binder is known to transform to silica,
carbon dioxide, water vapor and possibly some other carbona36
ceous materials.
The following equation is postulated to
describe the decomposition of the binder:
SiO(CH-.)^ + 40^ 7 1 SiO^ + 2C0^ + 3H_0
3 2
2 heat
2
2
2
(2)
Two tests were done to consider if Equation (2) is a
reasonable, qualitative statement of the decomposition of the
binder due to heat treatment.
The molecular weights of
SiO(CH-)^ and SiO^ are 74 g/mole and 60 g/mole, respectively.
If Equation (2) is correct, then the binder should transform
to silica after sufficient heat treatment.
In other words,
the weight of material in a crucible should change from 74
g/mole to 60 g/mole, a 19 w/o loss.
The binder was prepared in the following m.anner for
both tests.
The binder was dried in an aluminum foil containe:
68
in an electric dryer at 160°F for about three days and then
dried in air for about one week.
The dried binder was pulled
off the aluminum foil and kept in air for the two tests.
The first test considered long term heating of the
dried binder.
The binder (1.54 g) was put in a porcelain
crucible and heated in an electric furnace at 427°C for about
2 4 hours. After the heat treatment, the remaining material
weighed 1.24 g, for a 19.5 w/o loss.
In the second test, the binder was heated for short
periods of time typical of the sealing procedure.
The dried
binder was not brittle, so it was thermally treated in liquid
nitrogen to harden the resin, and then crushed and sized.
The size ranges were +12 mesh, -12 to +20 mesh, and -20 to
+35 mesh, and the following respective amounts of each size
were put in six porecelain crucibles:
0.243 g, and about 0.106 g.
averaged 0.470 g.
about 0.121 g, about
The total amount in each crucible
The sizing and weighing v/as done to get
approximately uniform binder surface area in each crucible.
The six crucibles were heated individually in an electric
kiln at 9 82°C for from one to six minutes.
The remaining
material was weighed and the percent weight losses as a
function of heating time are shown in Figure 20.
The results of these two tests show that about 2 0 w/o
material is lost during heat treatment of the binder, and this
amount is consistent with Equation (2). Other information to
consider is that water was observed dripping from the hollov/
69
1
2
3
4
5
6
Heating Time (min)
Figure 20. The Weight Loss of Binder Due to Heat
Treatment at 982^0.
70
stool that held the spark plugs during the sealing procedure.
The conclusion is that Equation (2) is an approximate
statement of the effect of heat treatment on the binder.
71
APPENDIX B
A DESCRIPTION OF THE SPARK PLUG INSULATOR,
ELECTRODES AND COPPER-GLASS
Spark plug insulators, terminal screws, and center
wires were obtained from a spark plug manufacturer, and they
are pictured in Figure 21.
The insulator is a high alumina
body with a small shoulder in the lower part of the bore on
which the center wire head rests when it is placed into the
bore.
The term.inal screw is C1008 steel with a nickel flash
coating.
The terminal screw has coaxial grooves on its lower
end so that a firm mechanical seal between the copper glass
and the grooves is created when the sealed plug cools.
center wire is made of inconel 600.
The
The center wire head is
shaped like a parallelepiped and has two grooves crimped in
it on both long sides.
The grooves and the shaped head pre-
vent the' center wire from turning in the sealed plug.
The copper glass was also obtained from a spark plug
manufacturer.
powder.
w/o
It consists of 4 0 v//o glass and 60 w/o copper
The composition of the glass is 69.9 w/o Si02/ 25.7
^2^3' ^'^ ^/° ^•'•2*^3 ^^^
^'^
^^^
Na20.
72
.t^^0^s:
Figure 21.
The Spark Plug Electrodes and Insulator
Showing the Shoulder on Which the Center
Wire Head Rests in the Lower Part of the
Bore.
73
APPENDIX C
ATTEMPTS TO REDUCE THE SCATTER IN THE
SPARK PLUG VALUES
The scatter in the spark plug resistance values as
shown by the standard deviations is not desireable.
The
scatter can be studied by considering the ratio o/X,
where
o is the standard deviation and X is the mean of a sample of
plugs.
Four tests were made to determine the cause of the
scatter and find a way to reduce it.
The tests were (a) to
correlate the resistance values with the distance between the
copper glass and the center wire (hereafter called the gap
distance), (b) to increase the B.C content, (c) to reduce
the particle size of the three dry components, and (d) to
consider different electrode geometries.
In order to determine if there was a correlation
between the resistance values and the gap distance, thirteen
spark plugs of the same composition were sectioned and
polished as described in Chapter III.
The resistance values
were measured before and after the sectioning process.
Photo-
micrographs v/ere made at 5Ox and the gap distance was measured at eleven equally spaced positions across the photomicrograph.
There was no correlation between the resistance
values and the gap distance as can be seen in the data in
Table 8.
These results show that the scatter in the resis-
74
Table 8.
The Lack of Correlation Between Resistance Values
and the Gap Distance.
Plug ID
R
:
^^
^RT
R
c
X
T
8
5.4
17.0
58
7.17
3
7.7
15.0
28
2.3
6
6.5
13.8
36
3.0
4
6.4
13.2
27
2.4
1
7.8
12.8
35
3.1
2
7.7
11.4
42
3.7
5
6.7
11.0
35
3.1
7
5.4
9.4
36
3.1
11
4.0
6.8
36
3.1
10
4.0
6.3
37
3.4
The room temperature resistance value of the sealed
plug {kn)
R :
The resistance value of the sealed plug after
sectioning {kQ)
X:
The average measurement of 11 distances (x ) across
the gap (millimeters at 50x)
T:
Since resistance is proportional to length and since
parallel resistors add inversely, then
1/T = 1/x, + I/X2 + 1/x^ + . . . + 1/x
75
tance values is not due to variable gap distance.
Similarly,
the results show that the resistance values are not linearlydependent on the gap distance.
This last result suggests
that the conduction path is not jusc across the gap but also
through the resistor material that is adjacent to the insulator walls.
A possible contribution to the scatter in the resistance values may be the low B.C content.
The B.C particles
may have been poorly dispersed and, if so, the resistor mixes
were inhomogeneous.
a
In order to possibly solve this problem,,
series of trial resistor mixes were formulated with the
B.C content varying from 2.7 to 3.6 w/o.
These amounts were
up to about three times as much B.C as was used in previous
trial compositions.
It has been shown in Table 7 that
increasing the B.C content lowers the resistance values.
Therefore, for this test, the Ti02 content was reduced in
order to balance the effect of the B C increase on the resistance values.
The compositions of the resistor mixes and the
resulting resistance values are given in Table 9,
There is
10 w/o binder in all the compositions and the amount of glass is
the balance for 100 w/o.
As the 3.C content was increased
for each amount of Ti02 (5, 6.5 and 8 w/o), the ratio a/X
had no consistent trend.
If anything, the scatter in the
resistance values increased instead of decreased as the B .C
content was increased.
The apparent result of this experiment
76
Table 9.
Mix ID
The Composition of the Resistor Mixes and the Resistance Values of the 3park Plugs Incorporating These
Mixes.
w/o B,C
w/o TiO^
a
n
a/X
U4
2.8
5
51
24
6
0.47
U7
3.2
5
36
14
5
0.39
UIO
3.6
5
18
12
4
0.67
U5
2.75
6.5
51
30
5
0.59
U8
3.14
6.5
27
17
6
0.63
Ull
3.53
6.5
35
44
5
1.3
U6
2.7
8
35
15
6
0.43
U9
3.09
8
8
3
5
0.41
U12
3.47
8
16
10
6
0.63
77
was that increasing the B.C content did not reduce the
scatter in the resistance values as measured by the ratio
a/X.
Some of the scatter in resistance values may have been
due to laminar flow, and this test vzas an attempt to reduce
the laminar flov/ by using a smaller particle size for the
glass, TiO^ and B.C.
tor mix T4.
The composition v/as the same as resis-
The three dry components were mixed by hand and
then ground and mixed in a Eleuler Mill (model 1193 LOS1659)
for one minute.
The binder, (2.50 g) was diluted with 9.70 g
of acetone to facilitate the dispersement of the binder in
the small size mix.
Seventeen plugs were sealed vjith this mix and the
average resistance value was 37.6 k^.
The standard deviation
was 17.1 so that the ratio a/X was 0.45, a high value.
The
reason for this behavior is not understood at this time.
Low resistance values were anticipated because reducing the
particle size was expected to create many more conduction
paths, thus reducing the resistance values.
Further, increasing
the number of conduction paths was expected to make the resistance values more uniform because a few highly conductive
and/or highly resistive paths would have less effect on the
overall resistance value.
Some of the scatter in the resistance values may have
been due to varaible bonding betv/een the electrodes and the
resistor mix and/or the copper glass.
The following changes
78
in the electrode geometries v/ere made in order to reduce this
effect.
The terminal screw was replaced by a steel cylinder
0.178 inches in diameter that closely matched with the bore
of the insulator resulting in a "snug" fit.
The center wire
was replaced by a nail with the head that also made a "snug"
fit in the bore, resting on the shoulder of the insulator, as
did the center wire.
Both electrode surfaces contacting the
resistor mix were dipped in light machine oil, and then in
pure copper powder prior to loading and sealing.
No copper
glass was used.
The standard volume of mix and sealing procedures were
used and four spark plugs were sealed using trial resistor mix
T4.
The average of the resistance values was 1.84 kfi, the
standard deviation was 0.334, and the ratio a/X was 0.18.
This ratio was the lowest value of any resistor mix in the T,
U, or V series, and the average resistance value was also low.
These two changes could be due to a significant reduction in
the contract resistance because of the larger surface area of
the electrodes.
These changes could also be due to a
significant modification of the conduction path through the
resistor m.aterial since the flow of the resistor material was
limited by the "snug" fit of both electrodes in the bore.
A microstructural examination of the sealed resistor
material in these plugs showed that the laminar flow was
altered.
Because of the difficulty in making the customized
electrodes, no attempt was made to determine how much of each
79
change (the decrease in the resistance values and the decrease
in the scatter of the resistance values) was due to the
increased contact area and to the modification in the laminar
flow.
Both changes v/ere due to the electrode geometry modifi-
cations.
80
APPENDIX D
A CALCULATION FOPv THE VOLU?^ OF TiO^
IN TEE RESISTOR MIX
The following calculation uses geometrical considerations to show that there is enough Ti02 ^^ ^^^ resistor mix
to form 25 layers of TiO^.
An average gap distance (the
distance from the copper glass to the center wire head)
from Table 8 is 37 mm at 50x or, actually, 750 wm.
If an
average glass layer is 30 ym then there should be 25 glass
layers between the copper glass and the center wire head.
About 25 glass-TiOp layers can be counted in the gap in the
photomicrographs in Figures 17 and 18 (note the overlap of
these two figures).
Suppose each layer is 3 mm in diameter and 3 ym thick.
-4
3
The volume of 25 layers is calculated to be 5 x 10
cm .
A resistor mix contains 10 w/o Ti02.
The bulk density of the
TiOp is 0.55 g/cc and 193 mg of resistor m.ix are in each spark
-'>
3
plug. Therefore, there is 3.5 x 10 " cm of TiO^ m each
spark plug.
This value is 70 times more than is needed to
form 25 layers of the dimensions used above.
These dimensions
are intended to be larger than actual layer dimensions.
Of
course, much of the Ti02 """^ """^ layers next to the insulator
walls.
This calculation was presented ro show that there is
enough
TiO^ in the resistor mix to create electrically con-
tinuous layers in the resistor mix.
81
BIBLIOGRi^PHY
1.
R. A. Webb, K. Schwartzwalder, and P. N. Keston,
"Resistor Composition," U. S. Patent 3,567,658, March 2,
1971.
2.
M. Oki, et al., "Resistor Built-in Spark Plug," U. S.
Patent 4,004,183, January 18, 1977.
3.
F. E. Heischinan, "Resistor Compositions and Spark Plugs
Having Integral Resistors," U. S. Patent 3,088,921,
May 7, 1963.
4.
R. R. Burgett, R. E. Massolle, and D. R. Van Uum,
"Relationship Between Spark Plugs and Engine-Radiated
Electromagnetic Interference," IE5S Transactions on
Electromagnetic Compatibility, VolT EMC-16, No. 3, p.
160, August 1974.
5.
Anonymous, "Reduced Ignition Interference," Electronics,
Vol. 22 (2), p. 118, July 1949.
6.
K. Schwartzwalder and A. S. Rulka, "Spark Plug Electrode
Seal," U. S. Patent 2,248,415, July 8, 1941; J. E. Blum
and J. W. Ezell, "Spark Plug with a Conductive Glass Seal
Electrode of Glass, Copper and Zinc," U. S. Patent
3,349,275, October 24, 1967.
7.
W. E. Counts, R. W. Smith, and K. Schwartzwalder,
"Resistor and Spark Plug Embodying Same," U. S. Patent
2,864,884, December 16, 1958.
8.
T. G. McDougal. K. Schwartzwalder, and A. S. Rulka,
"Resistor and Spark Plug Embodying Same," U. S. Patent
2.459,282, January 18, 1949.
9.
W. E. Counts, R. W. Smith, and K. Schwartzwalder,
"Resistor Composition and Devices Embodying Same,"
U. S. Patent 3,235,655, February 15, 1966.
10.
M. Yoshida and M. Sakai, "Self Sealable Glassy Resistor
Com.position for a Resistor Sealed Spark Plug," U. S.
Patent 4,006,106, February 1, 1977.
•
11.
G. Skinner, H. L. Johnston, and C. Beckett, Titanium
and Its Compounds, A Review of the Literature, H. L.
Johnston Enterprises, Columbus, 19 54.
82
12.
A. H. loffe. Physics of Semiconductors, Academic Press,
New York, 1960, (translated from the Russian).
13.
G. V. Samsonov, ed., The Oxide Handbook, IFI/Plenum,
New York, 1973, (translateid from'the Russian).
14.
H. K. Henisch, "Thermo-Electric and Conductive Properties
of Blue Titanium Dioxide," Blectrical Communication, Vol.
25, p. 163, 1948.
15.
J. H. Becker and W. R. Hosier, "Multiple-Band Conduction
in N-Type Rutile (Ti02)/" Physical Review, Vol. 137,
No. 6A, p. A1872, March 15, 1965.
16.
M. D. Earle, "The Electrical Conductivity of Titanium
Dioxide," Physical Review, Vol. 61, p. 56, 1942.
17.
H. H. Hausner, "Semiconducting Ceramic Materials,"
Journal of the American Ceramic Society, Vol. 3 0, No. 9,
p. 290, 1947.
18.
J. Takahashi, I. Yamai, and H. Saito, "Effect of Additives on the Sintering and the Electrical Conductivity
of Ti02," Yogyo-Kyokai-Shi, Vol. 83, No. 12, p. 589, 1975.
18b. R. G. Breckenridge and W. R. Hosier, "Electrical Properties of Titanium Dioxide Semiconductors," Physical Review,
Vol. 91, No. 4, p. 793, 1953.
19.
Per Kofstad, Nonstoichiometry, Diffusion and Electrical
Conductivity of Binary Metal Oxides, p. 137, John Wiley
and Sons, Inc., New York, 1972.
20.
R..F. Bartholomew and D. R. Frankl, "Electrical Properties
of Some Titanium Oxides," Physical Review, Vol. "187, No.
3, p. 828, 1969.
20b. A. L. Avdeev, Y. L. Danilyuk, and L. A. Rozenberg,
"Mechanism of Electrical Aging in Titanium Dioxide,"
Izvestiya Akademll Nauk SSSR, Neorganicheskie Materialy,
Vol. 8, No. 2, p. 263, 1972, (translated from the
Russian).
21.
O. A. Golikova, et. al., "Thermoelectric Properties of
Semiconducting Materials with a Nonstandard Conduction
Mechanism," Soviet Physics-Semiconductors, Vol. 6, No. 9,
p. 1488, 197T:
22.
R. R. Ridgv/ay, "Boron Carbide," Transactions of the
Electrochemical Society, Vol. 66, p. 117, 1934.
83
23.
T. Sugaya and 0. Watanabe^ "Electric Conduction of Boron
Carbide," Physica Status Golidi (A), Vol. 15, p. K181,
1973.
24.
V. S. Neshpor, V. P. Nikitin, and V. V. Rabotnov,
"Studying Certain Slectrical and Physical Properties of
Pyrolytic Boron Carbide," Soviet Powder Metallurgy and
Metal Ceramics, No. 11, (71), p. 889, 1968, (translated
from the Russian).
25.
L. M. Litz and R. A. Mercuri, "Oxidation of Boron
Carbide by Air, Water, and Air-Water Mixtures at Elevated Temperatures," Journal of the Electrochemical
Society, Vol. 110, No. 8, p. 921, 1963.
26.
T. N. Nazarchuk and L. N. Mekanoshina, "The Oxidation
of Boron Carbide," Soviet Powder Metallurgy and Metal
Ceram.ics, No. 2 (20), p. 123, 1964, (translated from
the Russian).
27.
D. A. Dominey, "Reaction of Boron Carbide with Oxygen
and the Effect of Preirradiation in a Nuclear Reactor,"
Chemical Society, London, Journal (A), p. 712, 1968,
Part I.
28.
G. W. Morey, The Properties of Glass, pp. 477-48 9.
Reinhold Publishing Corporation, New York, 1954.
29.
H. J. Van der Westhuitzen, W. C. A. Van Niekerk, and
W. J. de Wet, "Glass Resistance Measurements," Journal
of the American Ceramic Society, Vol. 58, No. 11-12,
1975.
30.
T. P. Dgebuadze and O. V. Mazurin, "Investigation of
the Electrical Properties of Low-Alkali Borosilicate
Glasses," Inorganic Materials, Vol. 4, Pt. 3, p. 1330,
1.968, (translated from the Russian).
31.
L. F. Oldfiald and R. A. Wright, "The Volatilization of
Constituents from Borosilicate Glass at Elevated Temperatures," Glass Technology, Vol. 3, No. 3, p. 59, 1962.
32.
P. G. Heslop, "The Properties of Borosilicate Glass,"
Laboratory Practice, Vol. 17, p. 1024, 1968.
33.
C, G, Wilson and A. C. Carter, "The Self-Diffusion of
Sodium Ions in a Borosilicate Glass and a Soda-Lime
Glass," Physics and Chemistry of Glasses, Vol. 5, No. 4,
p. Ill, 1964.
84
34.
J. R. Hut chins,, III and R. V. Ear ring ton, Glass,
reprinted from Kirk-Othiner, Encyclopedia of Chemcial
Technology, 2nd ed, , "i-'ol. 10, p. 533, John V7iley and
Sons, Inc., New vork, 196 6.
35.
Personal Communication with R. Feeney.
36.
Personal Communicatior! with B. Finzel.